Maximum Intensity Projections

[callan]

The rendering engine interface (and underlying code) should be extended to provide support for maximum intensity projections.

[callan]

Scheduling this after the thumbnail service initial work is done in iteration4.

[callan]

Plopping these in M4.

[jason]

Some comments and thoughts on this:

• Standard approach for calculation of projection is maximum intensity, but summed projects are sometimes used, so should have both.

• The calculation of a projection can be relatively easy -- few sections, or (almost infinitely) hard. First, a projection of a 3D image requires reading all the data, so is I/O and compute intensive-- MIP has an 'if' for every pixel x N sections.

• This facility must support timelapse, multi-channel images. 5D images > 1GB are increasingly common, so a 'projection' might really mean 300 projections of 256 x 256 x 20 x 2. Can we burden our rendering engine with that, while 5 other users hit it with other requests?

• It's not clear to me how to present the resulting calculation to the user:
  • in a new window?
  • in the same window, with a switch to back and forth between 3D data and projection?
  • as a TIFF or MPEG dumped to the user's client?

• For multi-channel images, must calculate projection for individual channels and overlays.

• How do we store the result of this calc:
  • as a TIFF attached to the 5D image?
  • as a new set of Pixels
  • as a new Image (probably not)

We really need to define how this will be done, and then decide if we can have the RE do this work, or if this is a next step for the Python system Donald is building (link?), or use something else?

[callan]

We're going to try and get an initial version of this into milestone:3.0-Beta3 with the help of the additions of #891 and a new projections service.

[jason]

Hey all-

What's the story here? We agreed this has to be in Beta3, and is clear our users need it.

Thanks!

[jason]

This from Alexia (email on May 13):

"I am still wondering why we can't do projections from our own computers, which do not usually have problems accessing Zeus. It will allow us to be more productive! I talked with Chris about that a long time ago (probably in November), but it doesn't seem to be a top priority. I know the programmers have a lot of work to do, but I really think that for users, this is a top priority..."

Let's set a date for this so users know what is going on; based on recent discussions, what about July 1? July 15?

[jburel]

We can always set a date as long as we don't specify the year ;-)

More seriously, July 15 seems reasonable. The more we wait, the more upset the users will become.

[callan]

Initial implementation in r2588.

[callan]

Full implementation including Blitz mappings, enumerations, services and full pixels set projections now available in r2752.

[callan]

Fixed missing Blitz ServiceFactory method in r2755 and ensured that correct enumerations are used for the Blitz interface.

[callan]

r2764 has now been tested with the RenderingEngine and IProjection interfaces from Blitz and JBoss, should now be feature complete.

[jburel]

I have an image with 4 channels. Project only one channel (index 2) The following error occurred:

java.lang.IndexOutOfBoundsException: Index: 2, Size: 1
        at java.util.ArrayList.RangeCheck(ArrayList.java:546)
        at java.util.ArrayList.get(ArrayList.java:321)
        at org.hibernate.collection.PersistentList.get(PersistentList.java:271)
        at ome.model.core.Pixels.getChannel(Pixels.java:1174)
        at ome.services.projection.ProjectionBean.projectPixels(ProjectionBean.java:273)

[jburel]

This time I select all the channels and I have the following error:

org.springframework.dao.InvalidDataAccessResourceUsageException: could not insert: [ome.model.stats.StatsInfo]; nested exception is org.hibernate.exception.DataException: could not insert: [ome.model.stats.StatsInfo]
        at org.springframework.orm.hibernate3.SessionFactoryUtils.convertHibernateAccessException(SessionFactoryUtils.java:618)
        at org.springframework.orm.hibernate3.HibernateAccessor.convertHibernateAccessException(HibernateAccessor.java:412)
        at org.springframework.orm.hibernate3.HibernateTemplate.doExecute(HibernateTemplate.java:424)
        at org.springframework.orm.hibernate3.HibernateTemplate.executeWithNativeSession(HibernateTemplate.java:374)
        at org.springframework.orm.hibernate3.HibernateTemplate.flush(HibernateTemplate.java:886)
... more

Caused by: org.postgresql.util.PSQLException: ERROR: type "double precision" value out of range: underflow
        at org.postgresql.core.v3.QueryExecutorImpl.receiveErrorResponse(QueryExecutorImpl.java:1512)
        at org.postgresql.core.v3.QueryExecutorImpl.processResults(QueryExecutorImpl.java:1297)
        at org.postgresql.core.v3.QueryExecutorImpl.execute(QueryExecutorImpl.java:188)
        at org.postgresql.jdbc2.AbstractJdbc2Statement.execute(AbstractJdbc2Statement.java:437)
        at org.postgresql.jdbc2.AbstractJdbc2Statement.executeWithFlags(AbstractJdbc2Statement.java:353)
        at org.postgresql.jdbc2.AbstractJdbc2Statement.executeUpdate(AbstractJdbc2Statement.java:307)
        at com.mchange.v2.c3p0.impl.NewProxyPreparedStatement.executeUpdate(NewProxyPreparedStatement.java:105)
        at org.hibernate.jdbc.NonBatchingBatcher.addToBatch(NonBatchingBatcher.java:23)
        at org.hibernate.persister.entity.AbstractEntityPersister.insert(AbstractEntityPersister.java:2252)

[callan]

r2782 should fix the channel index mismatch bug.